MaxScale/Documentation/Tutorials/MaxScale-Failover-with-Keepalived-and-MaxCtrl.md

MaxScale Failover with Keepalived and MaxCtrl

Introduction

Keepalived is a routing software for load balancing and high-availability. It has several applications, but for this tutorial the goal is to set up a simple IP failover between two servers running MaxScale. If the main server fails the backup machine takes over, receiving any new connections. The Keepalived settings used in this tutorial follow the example given in Simple keepalived failover setup on Ubuntu 14.04.

Two hosts and one client machine are used, all in the same LAN. Hosts run MaxScale and Keepalived. The backend servers may be running on one of the hosts, e.g. in docker containers, or on separate machines for a more realistic setup. Clients connect to the virtual IP (VIP), which is claimed by the current master host.

Once configured and running, the different Keepalived nodes continuously broadcast their status to the network and listen for each other. If a node does not receive a status message from another node with a higher priority than itself, it will claim the VIP, effectively becoming the master. Thus, a node can be put online or removed by starting and stopping the Keepalived service.

If the current master node is removed (e.g. by stopping the service or pulling the network cable) the remaining nodes will quickly elect a new master and future traffic to the VIP will be directed to that node. Any connections to the old master node will naturally break. If the old master comes back online, it will again claim the VIP, breaking any connections to the backup machine.

MaxScale has no knowledge of this even happening. Both MaxScales are running normally, monitoring the backend servers and listening for client connections. Since clients are connecting through the VIP, only the machine claiming the VIP will receive incoming connections. The connections between MaxScale and the backends are using real IPs and are unaffected by the VIP.

Configuration

MaxScale does not require any specific configuration to work with Keepalived in this simple setup, it just needs to be running on both hosts. The MaxScale configurations should be similar to the extent that both look identical to connecting clients. In practice the listening ports and related services should be the same. Setting the service-level setting “version_string” to different values on the MaxScale nodes is recommended, as it will be printed to any connecting clients indicating which node was connected to.

Keepalived requires specific setups on both machines. On the primary host, the /etc/keepalived/keepalived.conf-file should be as follows.

vrrp_instance VI_1 {
  state MASTER
  interface eth0
  virtual_router_id 51
  priority 150
  advert_int 1
  authentication {
    auth_type PASS
    auth_pass mypass
  }
  virtual_ipaddress {
    192.168.1.123
  }
}

The state must be MASTER on both hosts. virtual_router_id and auth_pass must be identical on all hosts. The interface defines the network interface used. This depends on the system, but often the correct value is eth0, enp0s12f3 or similar. priority defines the voting strength between different Keepalived instances when negotiating on which should be the master. The instances should have different values of priority. In this example, the backup host(s) could have priority 149, 148 and so on. advert_int is the interval between a host “advertising” its existence to other Keepalived host. One second is a reasonable value.

virtual_ipaddress (VIP) is the IP the different Keepalived hosts try to claim and must be identical between the hosts. For IP negotiation to work, the VIP must be in the local network address space and unclaimed by any other machine in the LAN. An example keepalived.conf-file for a backup host is listed below.

vrrp_instance VI_1 {
  state MASTER
  interface eth0
  virtual_router_id 51
  priority 100
  advert_int 1
  authentication {
    auth_type PASS
    auth_pass mypass
  }
  virtual_ipaddress {
    192.168.1.123
  }
}

Once the Keepalived service is running, recent log entries can be printed with the command service keepalived status.

Aug 11 10:27:59 maxscale2 Keepalived_vrrp[27369]: VRRP_Instance(VI_1) Received higher prio advert
Aug 11 10:27:59 maxscale2 Keepalived_vrrp[27369]: VRRP_Instance(VI_1) Entering BACKUP STATE
Aug 11 10:27:59 maxscale2 Keepalived_vrrp[27369]: VRRP_Instance(VI_1) removing protocol VIPs.

MaxScale health check

So far, none of this tutorial has been MaxScale-specific and the health of the MaxScale process has been ignored. To ensure that MaxScale is running on the current master host, a check script should be set. Keepalived runs the script regularly and if the script returns an error value, the Keepalived node will assume that it has failed, stops broadcasting its state and relinquishes the VIP. This allows another node to take the master status and claim the VIP. To define a check script, modify the configuration as follows. The example is for the primary node. See Keepalived Check and Notify Scripts for more information.

vrrp_script chk_myscript {
  script "/home/scripts/is_maxscale_running.sh"
  interval 2 # check every 2 seconds
  fall 2 # require 2 failures for KO
  rise 2 # require 2 successes for OK
}

vrrp_instance VI_1 {
  state MASTER
  interface wlp2s0
  virtual_router_id 51
  priority 150
  advert_int 1
  authentication {
    auth_type PASS
    auth_pass mypass
  }
  virtual_ipaddress {
    192.168.1.13
  }
  track_script {
    chk_myscript
  }
}

An example script, is_maxscale_running.sh, is listed below. The script uses MaxAdmin to try to contact the locally running MaxScale and request a server list, then check that the list has at least some expected elements. The timeout command ensures the MaxAdmin call exits in reasonable time. The script detects if MaxScale has crashed, is stuck or is totally overburdened and no longer responds to connections.

#!/bin/bash
fileName="maxadmin_output.txt"
rm $fileName
timeout 2s maxadmin list servers > $fileName
to_result=$?
if [ $to_result -ge 1 ]
then
  echo Timed out or error, timeout returned $to_result
  exit 3
else
  echo MaxAdmin success, rval is $to_result
  echo Checking maxadmin output sanity
  grep1=$(grep server1 $fileName)
  grep2=$(grep server2 $fileName)

  if [ "$grep1" ] && [ "$grep2" ]
  then
    echo All is fine
    exit 0
  else
    echo Something is wrong
    exit 3
  fi
fi

Aug 11 10:51:56 maxscale2 Keepalived_vrrp[20257]: VRRP_Script(chk_myscript) failed
Aug 11 10:51:57 maxscale2 Keepalived_vrrp[20257]: VRRP_Instance(VI_1) Entering FAULT STATE
Aug 11 10:51:57 maxscale2 Keepalived_vrrp[20257]: VRRP_Instance(VI_1) removing protocol VIPs.
Aug 11 10:51:57 maxscale2 Keepalived_vrrp[20257]: VRRP_Instance(VI_1) Now in FAULT state

MaxScale active/passive-setting

When using multiple MaxScales with replication cluster management features (failover, switchover, rejoin), only one MaxScale instance should be allowed to modify the cluster at any given time. This instance should be the one with MASTER Keepalived status. MaxScale itself does not know its state, but MaxCtrl (a replacement for MaxAdmin) can set a MaxScale instance to passive mode. As of version 2.2.2, a passive MaxScale behaves similar to an active one with the distinction that it won't perform failover, switchover or rejoin. Even manual versions of these commands will end in error. The passive/active mode differences may be expanded in the future.

To have Keepalived modify the MaxScale operating mode, a notify script is needed. This script is ran whenever Keepalived changes its state. The script file is defined in the Keepalived configuration file as notify.

...
virtual_ipaddress {
  192.168.1.13
}
track_script {
  chk_myscript
}
notify /home/user/notify_script.sh
...

Keepalived calls the script with three parameters. In our case, only the third parameter, STATE, is relevant. An example script is below.

#!/bin/bash

TYPE=$1
NAME=$2
STATE=$3

OUTFILE=/home/user/state.txt

case $STATE in
  "MASTER") echo "Setting this MaxScale node to active mode" > $OUTFILE
                  maxctrl alter maxscale passive false
                  exit 0
                  ;;
  "BACKUP") echo "Setting this MaxScale node to passive mode" > $OUTFILE
                  maxctrl alter maxscale passive true
                  exit 0
                  ;;
  "FAULT")  echo "MaxScale failed the status check." > $OUTFILE
                  maxctrl alter maxscale passive true
                  exit 0
                  ;;
        *)        echo "Unknown state" > $OUTFILE
                  exit 1
                  ;;
esac

The script logs the current state to a text file and sets the operating mode of MaxScale. The FAULT case also attempts to set MaxScale to passive mode, although the MaxCtrl command will likely fail.

If all MaxScale/Keepalived instances have a similar notify script, only one MaxScale should ever be in active mode.